Theories of agency discuss the possible ways of designing complex systems that display purposeful activity. Theories of agency have been researched in several disciplines (Brooks, 1986; Agre and Chapman, 1987; Suchman, 1987; Hendriks-Jansen, 1996; Johnston and Brennan, 1996; Agre and Horswill, 1997; Clancey, 1997), where two main positions are found – which we will call the ‘deliberative’ and the ‘situational’ theories of agency. The two theories have quite different modes of representation and action selection. In previous papers (Johnston and Milton, 2001; Johnston and Milton, 2002a; Lederman et al., 2003; Lederman et al., 2004) we have argued that information systems are purposeful, and that methodologies and tools used to build them should be analysed using theories of agency. However, existing approaches to computerised information system design and development are implicitly informed by the deliberative theory of agency. An approach different from present systems analysis and design methodologies is needed because many information systems fail in pressured routine environments, where we would argue that the situational theory of agency provides a better description of purposeful activity.

Although it is possible to design information systems for pressured routine environments using traditional methodologies, many of these systems are ineffective, inefficient or not accepted by people using them. They work technically but fail, in the context in which they are placed, to support the routine work adequately. We have reason to believe that in order to achieve greater success and acceptance of information systems in routine environments, we need a methodology that explicitly acknowledges the situatedness of socio-technical systems and their components, of which human actors and technical artefacts are examples.

Our long-term aim^[1] is to develop a situational information systems analysis and design methodology informed by the situational theory of agency. The first step is to establish an initial methodology to use in later stages of the research, which will employ action research. Consequently, the aim of this paper is to develop a first-cut of such a methodology based on concepts from the situational theory of agency and supplemented by our exploration of evolved manual situational systems. Although the methodology is intended for designing computerised systems, the specific focus of this paper is on learning lessons from existing manual situational systems so that the initial methodology, based on the situational theory of agency, can incorporate and generalise important features of situated systems that are currently in use and known to be effective.

The method we use in the paper is first to extract key concepts from the situational theory of agency as it is understood in robotics and discussed in other disciplines. Based on these concepts, we draw up a skeleton of a methodology for analysing systems. Following this, we examine several manual systems that have been either designed by users or evolved from practice, and that are both discussed in the literature and appear to be situational. We begin by establishing that the systems can be explained using the situational theory of agency. These systems are then used to understand how to apply concepts from the situational theory in practical systems before incorporating our experiences into the tentative methodology. We conclude by showing how we intend to refine the methodology.